[arin-ppml] Draft Policy ARIN-2017-5: Equalization of Assignment Registration requirements between IPv4 and IPv6 - updated 2017-07-21

[hostmaster at uneedus.com]

The first draft of my proposal was very conservative.  For v6 I proposed 
the two smallest possible subnet values be exempted from SWIP, which was 
/60 and /64.  I figured that this would be enough for 16 subnets, enough 
for IOT and/or guest,wired, and wireless networks on different segments.

Those in the know on this list suggested that I was setting my sights too 
low, and we quickly added values up to /56 to be exempted from SWIP and 
reached consensus on "more than a /56".  The bus networks I speak of are 
just an example of many networks I know of in the real world where a 
single public IPv4 address is used, or a /32, which has never required 
SWIP, and the thinking that simply adding v6 should not change this.

Because of SLAAC, the smallest v6 subnet used is /64.  However, unlike v4, 
the current rule requires this smallest size network be SWIP'ed.  Unlike 
your government network, in the real ISP world, 95-99 percent of the 
customers have only a single v4 address, and therefore most customers in 
the v4 world has never triggered the SWIP requirements.  Adding the 
minimum /64 for v6 now triggers 100% SWIP and all the associated labor, 
and this is what I seek to address with this draft.  I do not believe that 
these small size networks should trigger SWIP simply because they decided 
to do the right thing and add IPv6.

Others spoke that getting rid of the conserving v4 mentality is what is 
needed, and that even ARIN policy considers the default end user site 
should be a /48.  This is in fact what I have at my home, divided into 3 
/64's, which seems very wasteful in v4 mentality which would suggest that 
I should not require more than a /60.  Others here pointed out the 
benefits of all end sites being /48, which will allow automatic 
configuration in future devices, uniformity of all networks, and expansion 
into things using their own subnets that we may not even consider viable 
at this time.

While my conserving nature tells me I should be happy with a /56 or /60, I 
have been convinced by others in the know that the benefits described are 
worth the extra address space, which will be grown into in the future. 
These same voices speak with the desire for uniformity of network size and 
that we should have a policy of /48 for each end site. This third round of 
the draft is where we are at.  Quite a bit of consensus has been reached 
so far that the draft policy should not do anything to encourage operators 
to assign less than a /48 to each site.

Since the numbers work for you and the assignments are already in place, 
maybe the majority opinion of not less than a /48 per site will not work 
for you, and nothing in the draft would require you to change this.  In 
fact with my bus example, I have been actually considering assigning /60's 
out of a master /48 to reduce the amount of fees paid to ARIN, a site 
non-standard subnet size of /60.

In defense of the current proposal, do remember the principle of "your 
network, your rules".  If you require each district or university to be 
registered in SWIP in order to reduce your own abuse workload, go ahead 
and make SWIP a local requirement of receiving an assignment of your 
allocation. The draft assumes every end point is a /48 and therefore is an 
endpoint not requiring registration, a fact that is not true on your 
network. Based on your own statements, it sounds like you will be mostly 
SWIP'ing /48's for each district, and the districts will be assigning 
something smaller for its sites, such as a /56.

I note that the draft proposal is just a minimum standard, and does not 
forbid an operator from requiring additional SWIP registration above these 
requirements in order to receive space from your allocation.  If I were 
you, I would record a proper SWIP record of /48 for each district and call 
it a day, unless a district asks to subdelegate their space to more than 
one contact, in which case I would divide their assignment into however 
administrative abuse contacts they require and hope that number is a power 
of 2.  This takes care of all of your identifed needs with your smaller 
than /48 end site assignments, but allows the default policy values for 
SWIP which are in your case too large for your network that is already in 
place.  The draft is based on the recommendation of /48 per end user site, 
so that future networks are encouraged to follow the /48 per site rule.

Albert Erdmann
Network Administrator
Paradise On Line Inc.

On Thu, 27 Jul 2017, Richard J Letts wrote:

>
> On this thread we've gone from near-real-time update of bus GPS co-ordinates to suggesting allocating over 64 subnets per student for most of our school districts was a bad idea and we should have allocated more(!)
>
> Some stats for SY2017	# districts: 317; # districts <=100 students: 46  ;	# districts <=1000 students: 173 (including the 46)  ; # districts >= 10,000: 33 (source: http://www.k12.wa.us/DataAdmin/enrollment.aspx)
> Initial allocations have been around for more than 7 years. In 7+ years no school district has come back and asked for more or a larger allocation than a /48.
>
> I'm going to point out the current policy supports how we're swiping address space and it's up to you to persuade me your change is worthwhile.
>
> Richard
>
>
> _______________________________________________
> PPML
> You are receiving this message because you are subscribed to
> the ARIN Public Policy Mailing List (ARIN-PPML at arin.net).
> Unsubscribe or manage your mailing list subscription at:
> http://lists.arin.net/mailman/listinfo/arin-ppml
> Please contact info at arin.net if you experience any issues.
>